journal6 ›› 2007, Vol. 28 ›› Issue (2): 63-66.

• 物理与电子 • 上一篇    下一篇

光照变化条件下的人脸识别预处理

  

  1. (1.吉首大学物理科学与信息工程学院,湖南 吉首 416000;2.北京装甲兵工程学院装备指挥与管理系,北京 100072;3.南京理工大学计算机科学与技术学院,江苏 南京 210094)
  • 出版日期:2007-03-25 发布日期:2012-06-18
  • 作者简介:赵英男(1973-),女,辽宁盖州人,吉首大学物理科学与信息工程学院讲师,博士,主要从事图像处理、模式识别等研究.
  • 基金资助:

    湖南省教育厅科学研究项目(05C142)

Pre-Processing for Face Recognition Under Different Illumination Conditions

  1. (1.College of Physics Science & Information Engineering,Jishou University,Jishou 416000,Hunan China;2.Department of Equipment Command and Management,Academy of Armored Force Engineering,Beijing 100072,China;3.School ofComputer Science & Technology,Nanjing University of Science & Technology,Nanjing 210094,China)
  • Online:2007-03-25 Published:2012-06-18

摘要:光照变化条件下的人脸识别是模式识别领域最具挑战性的问题之一.笔者提出一种基于区域正则化模型(RRM)的预处理技术,即将人脸图像分成4个区域,在每个区域进行逐点线性变换.该技术简单易行,并能较好的削弱光照变化对人脸识别的影响.在Yale B人脸库上,与常用的直方图均衡化方法(HE)、基于区域的直方图均衡化方法(RHE)进行了对比分析,实验结果表明该预处理技术的有效性和可行性.

关键词: 光照变化, 直方图均衡化, 区域直方图均衡化, 图像正则化, 人脸识别

Abstract: Face recognition under different illumination conditions is one of the challenging problems in pattern recognition field.In this paper,a pre-processing technique based on Region Regulation Model (RRM) is proposed.It divides a face image into four parts,and in each part a linear transformation point-by-point is conducted.The technique is feasible,simple and tractable in computing.Moreover,it is very effective to weaken the aftereffect of varying illumination.Comparing with the popular techniques,such as Histogram Equalization (HE) and Region-based Histogram Equalization (RHE),experiments are performed on Yale B face image database.The results illustrate that this pre-processing technique is available and efficient.

Key words: illumination variation, histogram equalization, region-based histogram equalization, image regulation, face recognition

公众号 电子书橱 超星期刊 手机浏览 在线QQ